Travis CI has new infrastructure using the Google Compute Engine which has both
faster CPUs and more memory, and we've been encouraged to switch as it should
help our build times! The only downside currently, however, is that IPv6 is
disabled, causing a number of standard library tests to fail.
Consequently this commit tweaks our travis config in a few ways:
* ccache is disabled as it's not working on GCE just yet
* Docker is used to run tests inside which reportedly will get IPv6 working
* A system LLVM installation is used instead of building LLVM itself. This is
primarily done to reduce build times, but we want automation for this sort of
behavior anyway and we can extend this in the future with building from source
as well if needed.
* gcc-specific logic is removed as the docker image for Ubuntu gives us a
recent-enough gcc by default.
I just removed the num_cpus dependency (because we don't want that in there), using 4 threads instead.
I should add that Veedrac asked me to submit this here in his name.
Especially when documenting the use of `0`, since zero looks very
similar to `O` in fonts not meant for displaying code.
Other literal characters, traits, etc should also use code formatting.
This change makes this documentation more internally consistent.
Before this change, circled is the character I was using this documentation to find out about and that confused me when it wasn't immediately clear what character it was:
<img width="1013" alt="screen shot 2015-09-28 at 10 13 31 pm" src="https://cloud.githubusercontent.com/assets/193874/10154708/c70815fe-6638-11e5-9acc-57c73a524203.png">
After this change:
<img width="981" alt="screen shot 2015-09-28 at 11 26 35 pm" src="https://cloud.githubusercontent.com/assets/193874/10154710/ce73eeb2-6638-11e5-98f7-902f58679316.png">
I ran `make check-docs` and didn't break anything ⭐
Especially when documenting the use of `0`, since zero looks very
similar to `O` in fonts not meant for displaying code.
Other literal characters, traits, etc should also use code formatting.
This change makes this documentation more internally consistent.
This commit updates the compiler to not attempt to use jemalloc for platforms
where jemalloc is never enabled. Currently the compiler attempts to link in
jemalloc based on whether `--disable-jemalloc` was specified at build time for
the compiler itself, but this is only the right decision for the host target,
not for other targets.
This still leaves a hole open where a set of target libraries are downloaded
which were built with `--disable-jemalloc` and the compiler is unaware of that,
but this is a pretty rare case so it can always be fixed later.
By putting an "unreachable" instruction into the default arm of a switch
instruction we can let LLVM know that the match is exhaustive, allowing
for better optimizations.
For example, this match:
```rust
pub enum Enum {
One,
Two,
Three,
}
impl Enum {
pub fn get_disc(self) -> u8 {
match self {
Enum::One => 0,
Enum::Two => 1,
Enum::Three => 2,
}
}
}
```
Currently compiles to this on x86_64:
```asm
.cfi_startproc
movzbl %dil, %ecx
cmpl $1, %ecx
setne %al
testb %cl, %cl
je .LBB0_2
incb %al
movb %al, %dil
.LBB0_2:
movb %dil, %al
retq
.Lfunc_end0:
```
But with this change we get:
```asm
.cfi_startproc
movb %dil, %al
retq
.Lfunc_end0:
```
This was non-obvious to me: with no example, I assumed `Electron {}` and
didn't know what else to try when it didn't work. The correct form is
weird because it looks like you're assigning the struct name rather than
an instance of the struct.
r? @steveklabnik
the example for `find` was misleading in that it fails to mention the result is either `None` or `Some` containing only the first match. Further confusing the issue is the `println!` statement, "We got some numbers!"
As discussed in the referenced issues, this PR makes rustc emit `__imp_<symbol>` stubs for all public static data to ensure smooth linking in on `-windows-msvc` targets.
Resolves#26591, cc #27438
This commit updates the `MatchIndices` and `RMatchIndices` iterators to follow
the same pattern as the `chars` and `char_indices` iterators. The `matches`
iterator currently yield `&str` elements, so the `MatchIndices` iterator now
yields the index of the match as well as the `&str` that matched (instead of
start/end indexes).
cc #27743
This wasn't complete (you need a `./configure`), and it is already
documented well in the main README.
Also adds a reference to the books that this also generates.
This was non-obvious to me: with no example, I assumed `Electron {}` and
didn't know what else to try when it didn't work. The correct form is
weird because it looks like you're assigning the struct name rather than
an instance of the struct.
This adds a new target, `x86_64-rumprun-netbsd`, and related changes to `std`.
Rumprun is a unikernel platform that provides a POSIX-y interface. For the most part, rumprun uses NetBSD's libc and drivers, therefore `target_os` is `netbsd`. However, being a unikernel, rumprun does not support process management, signals or virtual memory, so related functions might fail at runtime. For this reason, stack guards are disabled in `std`.
To support conditional compilation, `target_env` is set to `rumprun`. Maybe `target_vendor` would be technically more fitting, but it doesn't seem to be worth the hassle.
Code for rumprun is always cross-compiled, it uses always static linking and needs a custom linker. The target makes use of the newly introduced `no_default_libs` flag, as the rumprun linker will otherwise not use the correct search path.
Because of type inference, duplicate obligations exist and cause duplicate
errors. To avoid this, only display the first error for each (predicate,span).
The inclusion of the span is somewhat bikesheddy, but *is* the more
conservative option (it does not remove some instability, as duplicate
obligations are ignored by `duplicate_set` under some inference conditions).
Fixes#28098
cc #21528 (is it a dupe?)
the example for `find` was misleading in that it fails to mention the result is either `None` or `Some` containing only the first match. Further confusing the issue is the `println!` statement, "We got some numbers!"
This PR removes random remaining `Ident`s outside of libsyntax and performs general cleanup
In particular, interfaces of `Name` and `Ident` are tidied up, `Name`s and `Ident`s being small `Copy` aggregates are always passed to functions by value, and `Ident`s are never used as keys in maps, because `Ident` comparisons are tricky.
Although this PR closes https://github.com/rust-lang/rust/issues/6993 there's still work related to it:
- `Name` can be made `NonZero` to compress numerous `Option<Name>`s and `Option<Ident>`s but it requires const unsafe functions.
- Implementation of `PartialEq` on `Ident` should be eliminated and replaced with explicit hygienic, non-hygienic or member-wise comparisons.
- Finally, large parts of AST can potentially be converted to `Name`s in the same way as HIR to clearly separate identifiers used in hygienic and non-hygienic contexts.
r? @nrc
For most parts, rumprun currently looks like NetBSD, as they share the same
libc and drivers. However, being a unikernel, rumprun does not support
process management, signals or virtual memory, so related functions
might fail at runtime. Stack guards are disabled exactly for this reason.
Code for rumprun is always cross-compiled, it uses always static
linking and needs a custom linker.
